The gilt long end takes out 6%, the term-premium shock goes global, and Australia pays for it without the AI offset

The US session — a 2002 high in the morning, a retreat by the bell

The shape of Thursday was a yield spike that failed. The 10-year touched roughly 5.34% intraday, its highest since 2002, early in the session on the back of the 10:00 ET data, and then retreated to close near 5.24%, about five basis points lower on the day. Equities rallied into that retreat: settled ETF closes at the 16:00 ET stamp were SPY +0.18%, DIA +0.00% and QQQ +0.30%, which on the verified 30 September cash bases gives S&P ≈7,665.5 and Dow ≈50,908. Both CNBC and Yahoo headlined the session as a comeback as yields fell and chip stocks gained.

The data was the hawkish part and the market looked through it. ISM manufacturing printed 54.5, a tenth of a point below August, for a twenty-third consecutive month of expansion — but the prices-paid index jumped 6.8 points to 77.9, with new orders +1.6 to 55.3 and backlogs +4.6 to 56.4. An S&P Global manufacturing PMI at 55.9 was described as the highest in more than four years. Initial claims fell to 197,000 against a ~200,000 consensus, with continuing claims at 1.70m, the lowest in three years, and construction spending rose 0.9% m/m against a flat consensus. That is a prices-paid jump of nearly seven points, a four-year high in manufacturing and a three-year low in continuing claims, all in one morning. Read here as the reason the long end spiked before supply-and-demand technicals pulled it back, rather than as a repricing of the Fed — October odds barely moved.

Eight-plus Fed speakers crossed the tape and only one said anything about rates. Vice Chair Jefferson, at Virginia's Darden School, said "inflation has been too high for too long" — above target for over five years — flagged upside inflation risks from geopolitical tension and energy shocks, described the labour market as stabilised and "near maximum employment" at 4.1% unemployment, and endorsed the September hike as necessary to keep expectations anchored. Governor Waller's appearance was not a monetary policy speech at all — it was about FRED and AI at the St. Louis Fed's FRED Con — so despite a hawkish lean attributed to him in previews, he cannot have moved pricing. Bowman spoke on regulation. ⚠ One carried item needs correcting: Cook's widely-quoted "inflation has been too high for too long" line is dated 30 September, not 1 October; no 1 October text was posted for her.

Named single-stock moves were thin and several circulating figures do not survive inspection. Micron rose 2.28% after its FQ4 beat, having been down about 1% at the open — the reversal is itself the semiconductor story. Banks lagged: Citigroup −1.9%, PNC −1.8%, Bank of America −1.4%. ⚠ A widely-served "Corteva −84%" is a corporate split, a mechanical price adjustment, and is not published here as a decline. ⚠ Reported moves of +21% in Accenture, +12% in Synopsys and +11% in Cognizant are single-source, carry no stated catalyst, and are implausible for three large-cap IT names on one day; they are withheld pending verification. A closing sector table and an advance/decline count could not be obtained for the session.

Europe — the gilt long end broke 6% and took the continent with it

This was the session's real event. The UK 30-year gilt yield reached 6.00% for the first time since 1998 and the 10-year 5.48%, the highest since July 2007. Equities followed the curve down: the Stoxx 600 fell 1.30% to 626.65, the DAX 1.03% to 24,939.35, the CAC 40 1.62% to 7,835.31, the FTSE 100 1.68% to 10,428.27, the IBEX 2.17% to 19,005.30, the FTSE MIB about 2.21% to ≈50,237 and the Euro Stoxx 50 1.49% to 6,175.45. Every one of those chain-ties exactly to a verified 30 September prior.

Banks were the epicentre, and they fell on a day yields rose — which is the tell. In Milan, Mediobanca −4.23%, UniCredit −4.07% to €79.53, Intesa −4.02%, Banca MPS −3.74%, Banco BPM −3.64%, BPER −3.55%; in London HSBC −3.7%, Lloyds −3.2%, Barclays −3.0%; in the Euro Stoxx 50, Bayer −5.15% was worst and Deutsche Bank −1.83%. Utilities, tobacco and consumer staples followed on rate sensitivity. The pockets of resilience were energy, on crude, and technology: SAP +1.84%, Infineon +1.52%, Fincantieri +2.7% on a BofA Buy initiation, Rolls-Royce +1.9%, and Games Workshop −6.5% on the other side.

France supplied the fiscal half of the story. The PLF 2027 went to the Conseil des ministres and the Élysée communiqué settles its contents: €43bn of new consolidation measures within a €54bn total, a 5.0% of GDP deficit in 2027 against a revised 5.4% for 2026, public debt at 119.3% of GDP in 2026, and the number that matters for the OAT — the debt interest charge rising from €79.2bn to €91.2bn, a €12bn jump that consumes roughly 28% of the entire €43bn consolidation. ⚠ One carried item is wrong: the communiqué says the income tax scale "sera revalorisé" — it will be uprated, not frozen, which is a cost to the Treasury rather than the fiscal-drag gain a freeze implies. The freezes appear instead to fall on the civil-service point d'indice, housing benefit and family allowances, which is reported but not verified here. The same morning the AFT sold €11.999bn of OATs, with the benchmark November 2036 line clearing at 4.93% on a 2.00 bid-to-cover and the 2048 at 5.40%.

The euro-area September flash HICP has still not printed. It is due today at 11:00 CEST, 19:00 AEST, and the carried framing needs a correction: consensus has re-anchored to 3.6% y/y, so the 3.6–3.7% this desk has been carrying is now consensus rather than a hawkish surprise, and the 3.2% carried as "consensus" was in fact the August outturn. The national flashes released 29–30 September all beat — Germany CPI and HICP both 3.3% (3.1% and 3.2% expected) with energy +14.9% y/y and core unchanged at 2.4%; France IPCH 3.4% against a national IPC of 3.0%; Italy NIC 4.2%, a three-year high, with IPCA 4.1%; Spain CPI 4.9% and HICP 5.0%, the highest since February 2023, with core 3.1%. Read as an energy shock rather than a broadening: German core did not move and Spanish core is 3.1%, while the headline acceleration tracks the Hormuz oil pass-through. Euro-area unemployment, released 1 October, was stable at 6.4%.

Asia — Micron on a half-shut tape, and the dispersion is the story

With Hong Kong and the mainland both shut, Tokyo, Seoul and Taipei priced Micron alone, and the ranking is the opposite of semiconductor index weight. The Nikkei 225 rose 3.35% to 68,986.97 — the largest single-market move in this edition — the KOSDAQ 4.48% to 894.29, the KOSPI 1.95% to 6,971.35, snapping a three-session losing streak, and the TAIEX only 0.86% to 48,353.49, despite being the most semiconductor-weighted index in the region. Korean names were the only prices obtainable in the complex: Samsung Electronics +2.79%, SK Hynix +3.21%, Samsung Electro-Mechanics +3.17%. Japan > Korea > Taiwan is the inverse of memory exposure, which reads as an equipment-and-AI-capex move amplified by SoftBank's weight in a price-weighted index rather than a memory-fundamentals repricing.

Two things qualify it. The TOPIX "lagged" badly as banks, insurers and trading houses fell — so the market narrowed rather than broadened — and a 1 October TOPIX level could not be obtained, which makes the Nikkei/TOPIX spread the most informative missing number in this edition. And nothing domestic supported the tape: the BoJ Tankan missed, with large manufacturers' DI at +24 against +25 expected (a sixth straight quarterly improvement, but a miss), large non-manufacturers going backwards from +37 to +35, and firms' one-year-ahead inflation expectations easing to 2.6% from 2.7%; August industrial production fell 1.7% m/m against +1.7% expected and retail sales rose 2.7% against 3.3%. Korea's rally was domestically funded — foreigners sold for a fourth consecutive session while institutions bought.

Australia went the other way, falling 1.99% to 8,614.40 on the global long end with no semiconductor beta to express, and New Zealand was nearly flat at 13,810.51, −0.17%. India was the regional laggard — Sensex −0.79% to 71,909.70, Nifty −0.88% to 22,421.95 — absorbing FII outflows and the yield shock with no AI offset either, and it carries that underperformance into today's Gandhi Jayanti closure. ⚠ A reported Korean September export print of +83.5% y/y to a record US$120.9bn with semiconductors above US$60bn is not published here: the magnitudes are roughly double and triple Korea's normal run-rate, the two carriers share one origin, and it must be checked against the MOTIE primary before use.

Commodities and crypto — three tankers struck in Hormuz

Crude's move had a specific cause. Three tankers — Al Ruwais, Mersin Prosperity and Sinbad, all Liberia-flagged — were struck by unknown projectiles in the Strait of Hormuz between Monday evening and Wednesday; one fire was extinguished and all crews are safe. No attribution has been established and none is offered here. Brent December settled ≈$102.60, +4.43%, and WTI November ≈$93.04, +2.90%, on three corroborating routes. The US is also sending a third carrier strike group to the region, Ukrainian strikes on Russian refineries have prompted Moscow to halt diesel exports, and Chinese refiners cancelled some October fuel cargoes. Against that, J.P. Morgan said on 1 October that "the Middle East's oil export arteries are flowing again."

⚠⚠ The Brent November contract expired on 30 September at a $103.50 settle, and December is the front month from 1 October. The roll gap at expiry was $5.47, so any naive day-on-day change across the two sessions on a rolling series is a roll artefact, not a price move. It also means the headline Brent–WTI spread of $9.56 on 1 October is not comparable to the $13.08 like-for-like November spread on 30 September; on relative performance the spread widened, because Brent outperformed WTI by 153bp. Do not report Brent–WTI as having narrowed. Gold spot closed $4,175.90, +0.48% at a 16:05 ET stamp, silver $60.86. Bitcoin was $84,743, +1.43% from a verified prior — a lagging risk asset on a day Brent rose 4.4%, and CoinDesk's own colour, that the rally "lost momentum as bond yields remained elevated", is the cleanest statement of the regime.

The ECB — a hawkish diagnosis on a dovish calendar, and the OAT is the part nobody is pricing

The two-edition blocker is cleared, and the cause was a misreading rather than a data gap: "29% hike" and "70% hold" are complements of the same distribution. The publishable number is a 22–29% probability of a 25bp hike to DFR 2.75% on 29 October, corroborated across a real-money prediction market, post-decision market pricing and a futures-derived model. Two sources were rejected with reasons — one whose implied post-meeting rate sat 48bp below its own stated starting point while reporting hike probabilities, and one whose ~60% figure traced back to a page already on this desk's discard list and which was the origin of yesterday's phantom high end.

The reason October is priced low is Schnabel, and it is a hawkish argument pointing at a dovish outcome. On 30 September she made the case that the jump in global yields may itself damp price pressures, letting inflation return to target more gradually — the bond selloff doing some of the ECB's work. Econostream's tone meter registered a sharp drop on 30 September, captioned as Schnabel driving it "as urgency about further hikes fades"; ⚠ the same feed also has her arguing for pre-emptive action the same day, which is a tension this desk flags rather than resolves. Lagarde says the energy shock may require further increases but sees no evidence of second-round wage effects and wants a "measured response"; Demarco says above-expected underlying inflation "could be grounds to act"; Kažimír counsels patience; Escrivá notes rates are not yet in restrictive territory, which frames further hikes as removing accommodation.

The under-priced item is Nagel. On 1 October the Bundesbank president said the Transmission Protection Instrument exists to secure effective policy transmission and price stability, not to respond to specific sovereign spreads — said on the day France's budget landed, its 10-year hit a 2002 high, its 30-year cleared an auction at 5.40% and OAT–Bund set a twelve-month high. Read as a pre-emptive refusal to name a trigger level. That is a signal about France, and it is the most under-priced sentence in this edition. Today's euro-area flash HICP at 19:00 AEST is the week's largest repricing risk — but note the asymmetry has flipped: consensus has re-anchored to 3.6%, so an in-line print is no longer hawkish and the risk is two-sided. Core is the live variable, and every national core print came in contained while headline screamed energy.

The Fed — a low-information meeting and a high-information month

At 36–39% for October and roughly 91% cumulative for something by December, the market has made the 28 October meeting the low-information event and December the one that carries the risk. The sell side agrees: Goldman, JPMorgan and Morgan Stanley all point to December, and the disagreement has moved to whether a hike happens at all. The 7 October minutes are the record of a 12–0 hike that now looks like it may have been the top, and they land one day before the 30-year reopening. ⚠ One correction to carry: the minutes are of the 15–16 September meeting, not "the 16 September meeting". ⚠ And the standing item stands: the Fed monitor has not produced a settled post-17:00 ET read for five consecutive sessions, checked four times this run at the same 09:25 ET stamp. The 36–39% band from three independent routes is what this edition publishes instead, and it is firmer than any single page.

The ASX had its worst session of the week and it was a pure rates event. The S&P/ASX 200 closed 8,614.40, down 174.90 points or 1.99%, chain-tying exactly on three sources agreeing to the cent, for a weekly fall of 0.58% against the 25 September close of 8,665.00. The sector table is the whole story: real estate −3.37%, financials −2.13%, materials −1.89%, industrials −1.75%, consumer discretionary −1.18%, utilities −0.91%, telecoms −0.85%, information technology −0.62%. Real estate, discretionary and telecoms were the three sectors that led Wednesday's +0.92% CPI-relief bounce at +3.7%, +2.6% and +2.3% — they handed the entire move back in one session. Note also that materials did not lead the fall despite China being shut, and IT was the best sector while still falling: the ASX traded the bond market, not China, and it had nothing with which to express the global AI bid. Movers at the close, corroborated to the cent across two sources: Data#3 $12.63 +13.68% was the standout gainer, with PDI Gold +4.10% and 4DMedical +2.59%; against Liontown $0.79 −15.05% at a 52-week low, Lynas $12.64 −8.60% on its A$968m all-share acquisition of Meteoric Resources, IperionX −8.37% and Vulcan Energy −8.36%, with Rio Tinto −2.44%. ⚠ Close-basis breadth could not be obtained — 185 of 200 constituents were lower at 14:05, which is intraday — and neither turnover nor any dated SPI quote was available, the fifth consecutive edition without a futures indication.

The important Australian number this morning is not the index, it is the pricing: November is back to 36% from the ~20% carried, and the probability is internally consistent with a +9.1bp implied move. The entire dovish interpretation of the August monthly CPI has unwound inside two sessions. The bonds were ahead of it and say more: the ACGB curve bear-steepened +2 to +5bp to 2y 5.00, 3y 4.95, 10y 5.39 — on level differences, because the vendor's change column contradicted itself at all three tenors for the eleventh consecutive edition — putting 35bp over the 4.60% cash rate at the 3-year, about 1.4 further hikes, and 40bp at the 2-year, with 3s10s steepening to 44bp from 42. ⭐ And the Australian 10-year at 5.39% now trades above the US 10-year at roughly 5.24%: Australia is paying a yield premium to the United States at ten years. The bank calls split two-two and the hawks moved: Westpac has November as its base case at 4.85%, with chief economist Luci Ellis saying "the bar for a follow-up hike in November is low", joined by ANZ at 4.85%, against CBA (hold, risks tilted to a hike) and NAB (hold, peak 4.60%). Westpac adds a qualifier worth carrying: "a much higher bar for hikes beyond November", contingent on energy costs persisting. ⚠ Westpac's own website is a stale-source trap — its 9 September page still forecasts "November to 4.60%", which predates the 29 September hike.

The Bank's own words were unusually direct, and they were unanimous. The 29 September statement said "inflation remains elevated and some of the upside risks flagged in August are materialising" and that "recent inflation outcomes in Australia were stronger than expected", naming three drivers: Middle East conflict energy prices, AI-related technology costs and domestic capacity pressures. The labour market has eased as expected; consumer spending growth is moderating while business investment remains robust; guidance is to raise "if needed". A unanimous fourth hike of 2026 with upside risks described as materialising is a Board that has not finished — that is a reading of the language, not a forecast. ⭐ The diary is current (stamped 1 October, with no recurrence of the three-month staleness seen earlier in September) and it proves a negative: there has been no RBA speech since 29 September and none is scheduled until 19 October, when Christopher Kent speaks at the CBA Global Markets Conference, followed by Ellis Connolly on the 20th, Andrew Hauser on the 21st and Brad Jones on the 26th. September Board minutes publish 13 October; the Annual Report 15 October.

Thursday's domestic data was dovish at the margin and nobody looked at it. Job vacancies fell to 325,000, −0.9% q/q and −1.3% y/y — the clearest easing signal on the board — and the August trade surplus collapsed to $495m, down $856m from July, as imports rose 5.8% against exports at 3.7%: a $495m surplus is very narrow by Australian standards and the surplus is nearly gone. Building approvals fell 6.1% in August. ⚠ On the monthly CPI, the series naming matters and must be stated every time: the ABS publishes both an original m/m of 0.4% and a seasonally adjusted m/m of 0.7%, while the 0.2% that drove the dovish reading is the trimmed mean — three different numbers for one month. Ahead: building approvals 7 October, Labour Force 15 October, and the quarterly CPI on Wednesday 28 October, which is the resolution date and which lands after the November Board's data set is effectively fixed.

The household channel tightens on a date, and it is nine days away. CBA, NAB, Westpac and ANZ have all confirmed the full 25bp pass-through effective 9 October, with Macquarie following on 15 October and also lifting transaction and savings rates by 25bp — the only deposit-side move found. Headlines describe variable rates clearing 6% and the end of sub-6% mortgages. ⚠ The ~$79 a month on a $500,000 loan carried in previous editions is not in the source it was attributed to and is withdrawn as unsourced. That repricing lands into a housing market already falling: Cotality's September Home Value Index, verified against the primary, has national values −1.1% m/m, a sixth consecutive monthly fall and −5.2% from the March peak, with Brisbane −1.5%, Sydney −1.4%, Melbourne −0.7% and Darwin the only capital to rise at +0.4%. The annual spread is a two-speed market — Darwin +11.9% and Perth +10.1% against Sydney −7.0% and Melbourne −6.2% — and the breadth figure is the alarming one: 97% of capital-city suburbs fell over the three months to end-September. ⚠ On the suspiciously identical Sydney figure: the primary does print −1.4% for September, so it is not a vendor carry-forward; the coincidence with August remains unexplained rather than disproven. ⚠ And auction clearances need recomputing, as they do every week: the published national rate for the weekend of 26 September was 49.2%, but its own city rows — Sydney 56.2% of 1,029, Melbourne 59.5% of 291, Brisbane 37.1%, Adelaide 40.8%, Canberra 52.5% — recompute to 54.1%, a 4.9pp gap. Use 54.1%. ⚠⚠ Critically, that weekend PREDATES the hike. The first post-hike auction weekend is 3–4 October, tomorrow, so no clearance evidence of the 4.60% cash rate exists anywhere yet.

The currency did not reward the hawkish repricing, and that is the cost shock showing through. ⭐ AUD/USD is settled from the official source this edition: the RBA publishes AUD/USD and AUD/NZD in one 4pm AEST table, which ends the cross-computation problem permanently. The Aussie closed 0.6949, −0.40%, a fresh low below the carried nine-week trough, with the TWI at 64.6 — so this is trade-weighted weakness, not merely a strong dollar, and AUD/USD fell 0.40% on the RBA basis against only 0.06% on the New York close, meaning the Aussie fell through the Sydney afternoon and stabilised overnight. ⭐ The table also vindicates the carried prior: the RBA's 30 September print is 0.6977 against our 0.6976 — one pip — and it confirms the 0.7018 correction, since the RBA has 29 September at 0.6987. ⭐⭐ And it settles the AUD/NZD dispute outright at 1.2365, quoted directly rather than derived: the contested 1.2345–1.2388 band was two different bases on two different dates, not a disagreement. The cross rose 0.11% while the Aussie fell against the dollar, which is the 185bp RBA–RBNZ policy spread doing the work. ⚠ One source quoted the Aussie at 69.53 US cents "+0.14%"; against the RBA's own prior the arithmetic adjudicates the direction as down.

Iron ore and the China link are severed for ten days, and the miners outperformed into the dark. The benchmark's last print is $96.59/t on 30 September, with the Dalian January contract at 702.5 yuan — and both Dalian and the mainland are shut 1–7 October, so these are genuinely last-traded levels rather than stale carries. ⚠⚠ Singapore futures are the only live iron-ore price this week and no dated 1 October SGX print was obtainable, which is the most important commodity gap in this edition. Chinese crude steel output was −3.7% y/y in August with mill profitability constrained. The observation worth acting on: with no mainland price discovery until 8 October, the AUD and SGX are the only live China proxies, and Australian materials fell only 1.89% in a −1.99% tape — the miners outperformed with China dark. That is positioning rather than information, and it sets up a gap when Connect reopens on 8 October.

New Zealand is the regional expression of not having a central bank still hiking. The NZX 50 closed 13,810.51, −0.17%, chain-tying exactly — against Australia's −1.99%, on the same global yield shock, because the index is less rate-sensitive and the RBNZ sits 185bp lower with no tightening bias. ⚠ A 15:00 NZT read had it down 1.1%, implying a ~0.9pp recovery into the closing auction that could not be corroborated. The RBNZ's 28 October meeting is a Review rather than a full MPS, priced at roughly 11% for a hike — ⚠ carried and not re-verified against the Bank this edition — while the NZ 2-year at 3.92% against a 2.75% OCR embeds 117bp of tightening, which is a market pricing a cycle the Bank has not signalled. ⚠ The carried ANZ business confidence reading at a three-month low could not be re-verified.

THE ITEM THAT MATTERS MOST: a trigger written two editions ago fired exactly where it was written, and it was executed without discretion. V017's confirmation required "CCC through 1,150bp with IG flat". On 30 September CCC printed 1,179bp and IG printed 84bp, exactly flat — zero basis points, ending a six-session widening run that had gone +2bp then +1bp. The view confirms and conviction goes Med → High. Yesterday this desk declined to take a +1bp print as "flat" because the margin sat inside the series' own display rounding; today a genuine zero resolves it in the same direction the pre-commitment specified. The discipline that cost nothing yesterday paid today, and the diagnostic moved with it: the CCC/IG ratio broke 14× for the first time in the run.

THE SECOND ITEM IS A HEDGE PREMISE THAT INVERTED, AND IT STRENGTHENS RATHER THAN WEAKENS THE TRADE. The CBOE SKEW was re-sourced at 141.9 against the 146.3 carried — the one-year percentile has fallen from the 90th–96th band to the 28th. The published argument that deep protection is expensive is withdrawn. But put/call's nine-day average sits at the 8th percentile while AAII has bears at 46.5%: protection is cheap and under-owned. That is a better case for V028's expression than the one that broke, and it is stated as a replacement rather than a resize.

THE THIRD IS A ONE-BASIS-POINT MISS, FOR THE SECOND EDITION RUNNING, IN THE DIRECTION THAT COSTS MONEY. The pre-committed override on the 5s30s steepener opens it at 56bp or above on the official par curve. The 30 September official close was 55bp. It does not fire. A vendor curve implies 1 October at 60bp — five basis points through — but that vendor contradicts two others on the 10-year and the official 1 October curve does not exist yet. Per the convention written on 1 October, this desk does not round its own trigger in its own favour, and that cuts both ways: yesterday it meant not confirming a view, today it means not opening one that may well have paid.

Friday scorecard — and what this week got right and wrong

Seven views open, twenty-two closed to date, none closed this week. The trailing record is unchanged at 3 right / 12 wrong / 7 scratch across 22 decided — a hit rate of 3 of 15 decided views, which is bad and is stated plainly because the ledger is the accountability mechanism rather than the marketing.

What worked this week. V017 is the week's result: a trigger pre-committed in writing two editions ago, declined on a one-basis-point technicality on Thursday, and confirmed on Friday by an exact zero — conviction to High without a discretionary judgement anywhere in the chain. V025 is the week's P&L: +23.1bp and a twelve-month wide, with its fiscal catalyst now on the record and the mechanism quantified at €12bn of additional debt service. V006 is working at +4.35% with its primary rationale restored, and V023's marking problem is solved permanently by finding the official source rather than by arguing about legs. V028 is working at ≈−1.01% and the terms of its expression improved.

What went wrong, and three of the four were our own doing. (1) The 5s30s override missed by one basis point for the second consecutive edition, in the direction that costs money — four declines of this re-entry, two of which have now cost money, and a vendor curve says Thursday would have fired it. (2) Three carried Asian equity priors were wrong and all three were corrected by arithmetic this edition — the Nikkei, the TOPIX and the TAIEX — which is the third consecutive edition in which this desk's own inputs were a larger error source than its sources. (3) A load-bearing hedge input was three sessions stale: the SKEW at 146.3 when it was 141.9, and the percentile reading that followed from it was wrong by sixty points. (4) V024's reference contract expired underneath the view and the trigger had to be restated — a foreseeable calendar event that should have been written into the view when it was opened.

Portfolio-level read. The book is now concentrated in one idea expressed four ways: that the term premium is the binding constraint and the policy path is not. V025 is that trade in European sovereign credit, V029 in the US front end as the fade of a dovish repricing, V006 in an equity market with no offset to it, and V017 in the lowest tier of credit. The concentration is the risk: a soft payroll print tonight that rallies the long end would hurt V029 directly, relieve V006, and test whether V017's one-session separation was real. V028 and V023 are the two genuinely independent legs. Conviction rose twice this edition and both increases were argued from the view's own primary leg rather than from the price — V017 on its trigger, V006 on RBA pricing — which is the convention of 28 September working as intended.

These are analytical house views for a professional reader, not personalised financial advice. They carry explicit invalidation conditions and are logged and scored in the project ledger; conviction describes the asymmetry of the evidence, not the size of the expected move.